CN101772872B - Estimating Time Offset Between Static Clocks - Google Patents

Abstract

The present invention is concerned with an improved time synchronisation of two stationary clocks. A global time signal from a global time reference or common time source is used to calculate a common view based clock offset between two stationary clocks. In parallel, a network based clock offset between the two clocks is calculated based on messages exchanged over a communication network interconnecting the two clocks and without reverting to the global time reference. The two most recent values of the common view and network based clock offsets are then combined or superposed in a seamless or hitless way to produce a final time offset estimate. The combination of the independently calculated common view and network based clock offsets is a weighted average of the two values, involving respective weights based on quality estimates of the latter. Preferably, the time synchronization schemes based on the Global Positioning System (GPS) and a wide area communication network are combined in order to synchronize the stationary clocks of the Phasor Measurement Units (PMUs) of a Wide Area Monitoring System to a central server clock at the Network Control Center (NCC) of the system.

Description

Estimating Time Offset Between Static Clocks

technical field

The invention relates to the field of time synchronization between two geographically separated static clocks, in particular clocks of phasor measurement units for wide-area monitoring systems of power transmission networks.

Background technique

For wide-area monitoring of the transmission network, phasor measurement units (PMUs) are installed at distributed locations. The PMU performs sampling of current and voltage waveforms, calculates phasor values from the sampled waveforms, and periodically transmits the phasor values to a network control center (NCC) through a wide area communication network. The NCC monitors the status of the transmission network by comparing synchrophasor measurements received from distributed locations. Therefore, the synchronization of the phasor measurements is critical and requires synchronization of the sampling clocks of the PMUs. In order to make the system robust to transmission delays and jitter on the communication network, the phasor messages transmitted by the PMU include a time stamp indicating the exact time of measurement. Likewise, routers and switches in wide area communication networks require a similar degree of time synchronization.

Today, wide-area synchronization of distributed PMU clocks is performed using commercial Global Positioning System (GPS) time receivers. However, it is known that propagation and interference problems can exacerbate or even prevent GPS reception. The surrounding landscape may obscure a particular location from view to GPS satellites, or the solar wind may affect GPS signal reception for minutes. Although a navigating vehicle can easily switch to other systems for determining its position, no such alternative has been implemented for time synchronization of static clocks today.

Contents of the invention

It is therefore an object of the invention to improve the time synchronization of two static clocks. This object is achieved by a method and a device for estimating time offsets according to claims 1 and 7 . Other preferred embodiments are evident from the dependent patent claims.

According to the present invention, a global time signal from a global time reference or a time source in common view is used to calculate the common-view based clock offset between two static clocks instead of each clock with the global time reference between the two corresponding clock skews. In parallel, a network-based clock offset between two clocks is calculated based on messages exchanged over a communication network interconnecting the two clocks and without reverting to a global time reference. The two latest values of common-view and network-based clock bias are then combined or superimposed in a seamless or collision-free manner to produce the final time offset estimate.

In a preferred variant of the invention, the independently calculated combination of common-view and network-based clock offsets is a weighted average of said two values involving a quality estimate based on said two values corresponding weights. In an advantageous embodiment of the invention, the calculations of the common-view-based clock bias and the network-based clock bias are updated independently of each other and repeated with appropriate frequency.

In order to combine in an optimal way a time synchronization scheme based on the Global Positioning System (GPS) and a communication network for a static clock of a phasor measurement unit (PMU) of a wide-area monitoring system, the PMU client clock is synchronized to the system's network A central server clock at the Control Center (NCC) instead of being synchronized to the GPS clock itself. In practice, since GPS one-way time distribution (if available and functional) is expected to have higher accuracy than network-based synchronization, the method of the present invention will primarily act as a dynamic fallback to use the network whenever GPS synchronization fails plan. Although GPS is available, this approach is used to improve the accuracy of network-based synchronization, specifically by correcting for transmission jitter and delay asymmetry in network-based synchronization schemes.

Description of drawings

In the following, the subject-matter of the invention will be explained in more detail with reference to a preferred exemplary embodiment shown in the accompanying drawings, in which:

Figure 1 schematically shows a wide area monitoring system;

Figure 2 depicts the basic clock relationships; and

Fig. 3 shows a message sequence diagram according to the present invention.

The reference signs used in the drawings and their meanings are listed in summary form in the list of reference signs. In principle, identical parts have the same reference signs in the figures.

Detailed ways

Figure 1 shows a wide area monitoring system for a power transmission network, where several phasor measurement units (PMUs) are installed at distributed locations. The PMU calculates phasor values and will periodically send these phasor values to a Network Control Center (NCC) over a wide area communication network. Global Positioning System (GPS) satellites broadcast a global time signal. Any clock at one of the PMUs, ie client clock or slave clock C needs to be synchronized with the clock at the NCC, ie server clock or master clock S.

Figure 2 depicts the clock relationship. The clock C(t) of the PMU is determined by

C(t)＝ _φC ·t+ _θC (1)

, where θ _C is the time offset, φ _C ·t is the clock drift, and t represents the real time. Similarly, for the clock S(t) of the NCC,

S(t)＝ _φS t+ _θS (2)

The PMU's clock C(t) must be synchronized to the NCC's clock S(t), that is, the time offset x(t) of the PMU clock from the NCC clock must be estimated and then corrected at the client side, where

Here, the term y(t) represents the frequency deviation. The actual way to get that bias is:

1. One-way GPS measurement: Both client and server receive time information G(t) broadcasts from a common source, in practice from GPS satellites. The reception times measured by the client and the server are C'(t) and S'(t), respectively, for which the following holds:

C'(t)=G(t)+x _CG (t)+d _GC (4)

S'(t)=G(t)+ _xSG (t)+ _dGS (5)

In (4), x _CG is the offset between the client and the GPS clock, and d _GC is the propagation delay between the GPS satellites and the client. A similar definition is used in (5). Given the relative position of the clock and satellite, the delay can be compensated for, giving a corrected clock

C(t)=C'(t)-d _GC (6)

S(t)=S'(t)-d _GS (7)

By comparing the values C(t) and G(t), then x _CG (t)==C(t)-G(t) and x _SG (t)=S(t)-G(t) are directly obtained. This is the usual way for synchronizing PMU clocks. Signal x _CG (t) controls a local PMU oscillator that generates a 1 pps (1 pulse per second) and eg 10 MHz clock signal to synchronize PMU sampling and time stamping.

2. Common-view GPS measurements: In many applications it is not necessary or desired to estimate _xCG (t) and _xSG (t) separately, but obtaining the clock offset x(t) between client and server may be sufficient . This can be achieved by exploiting the fact that the GPS broadcast signal G(t) is co-viewed by them. For an agreed GPS broadcast time G(t _i ), the client and server record the reception times C(t _i ) and S(t _i ), and exchange these measurements in a non-time-critical manner over some communication network. Using (4) to (7) and Figure 2, the difference of these measurements is:

C(t _i )-S(t _i )＝x _CG (t _i )-x _SG (t _i )＝x _G (t _i ) (8)

Here, x _G (ti) denotes the deviation x(t) of C(t) relative to S(t), which is determined by offsetting the common GPS measurement value G(t _i ). A Kalman filter or other averaging techniques can be used to further improve the estimate of x(t). This common view method is the standard method used today to compare atomic clocks to define Coordinated Universal Time (UTC).

3. Bi-directional measurement: Clock client and server perform time measurement and exchange time measurement values in real time over a (time-critical) communication network to directly determine the time offset of the client. Standard two-way time synchronization protocols are SNTP on the Internet, and IEEE 1588 for devices connected to a LAN. In the case of using IEEE 1588 terminology, the basic steps are as follows:

(i) At time t _n , the server broadcasts a Sync message with timestamp S ₁ (tn), which is received by the client at C ₁ (tn). Considering the message transmission delay d _SC (t) of the Sync message from the server to the client, the following formula holds:

C ₁ (t _n )=S ₁ (t _n )+x(t _n )+d _SC (t _n ) (9)

where x(t) is the bias to be determined by this two-way method.

(ii) At client time C ₂ (t _n ), the client sends a Delay-Request message to the server, and the Delay-Request message is received by the server at time S ₂ (t _n ). The server responds with a Delay-Response message containing the value of S ₂ (t _n ). Similar to the above, d _CS (t) represents the propagation delay of the Delay-Request in the reverse direction from the client to the server.

S ₂ (t _n )=C ₂ (t _n )-x(t _n )+d _CS (t _n ) (10)

(iii) Now, 4 measurements S ₁ (t _n ), C ₁ (t _n ), C ₂ (t _n ) and S ₂ (t _n ) are available at the client. Assuming that the transmission delays are equal, that is, d _SC (t)=d _CS (t)=d(t), the client can solve (9) and (10) to obtain x(t _n ) as an estimate of the clock skew:

$\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; S</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span><span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; S</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 11</span>}<span\; class="notranslate">\; )</span>$

$x_T\left(t_{\mathrm{no}}\right)=(C_1\left(t_{\mathrm{no}}\right)-S_1\left(t_{\mathrm{no}}\right))-d\left(t_{\mathrm{no}}\right)=\frac{(C_1\left(t_{\mathrm{no}}\right)-S_1\left(t_{\mathrm{no}}\right))-(S_2\left(t_{\mathrm{no}}\right)-C_2\left(t_{\mathrm{no}}\right))}{2}$ (12)

where x _T (t _n ) denotes an estimate of the true deviation x(t) obtained at time t _n by the two-way measurement method. Methods such _as Kalman filters and averaging can _further improve the analysis of time and frequency deviations (x and _y ) of the estimation accuracy.

The two-way method for time synchronization relies on a communication network between the client and the server. The communication is time critical in the sense that any random variation and asymmetry in the delays d _SC (t) and d _CS (t) will affect the synchronization accuracy.

The detailed steps of the procedure are described below for a specific PMU client clock node C with a clock C(t). All PMUs execute this program in parallel to synchronize their individual clocks to the central server clock S(t) of server S. S is usually located at the NCC.

Fig. 3 shows the message sequence diagram of one round of the proposed procedure.

1. The server clock S(t) is initially free running (uncontrolled oscillator).

2. The client clock C(t) initially oscillates freely.

3. If GPS one-way time reception is available at server S, it receives GPS time G(t _i ) at time t _i and records its reception time S(t _i ). It can use this GPS time to control its oscillator.

4. The server S periodically broadcasts a Sync message to all clients Cs at time t _k . The message contains:

- the timestamp S ₁ (t _k ) of the message transmission,

- (if available from 3) the reception time S(t _i ) of the GPS time message.

5. C receives the Sync message and also records the message reception time C ₁ (t _k ).

6. If GPS one-way time reception is available at client C, it receives GPS time G(t _i ) at time t _i and records its reception time C(t _i ). If available from 4, C calculates its clock offset from S according to the common-view method,

x _G (t _i )＝C(t _i )-S(t _i )

C should use continuous measurements or other system information (e.g. information about satellite health and clock quality from GPS data) to determine the ^quality of the clock bias estimate _xG (t _i ), expressed e.g. by the variance _σG2 .

7. C sends the Delay-Request message to S. The message contains:

- the time stamp C ₂ (t _n ) of the message transmission,

And can be combined with periodic PMU phasor data messages in order to reduce the number of messages and message overhead on the communication network.

8. S receives the Delay-Request message and records the message reception time S ₂ (t _n ), and responds to C (at time t _n ′) with a Delay-Response message. The message contains:

- the value of S ₂ (t _n ),

- Optionally, its own transmitted timestamp S ₁ (t _n ').

9. C receives the Delay-Request message, and if it has received S ₁ (t _n ') at 8, it can record the message reception time C ₁ (t _n ').

10. C computes its clock offset from S measured by the two-way method as follows:

${\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; S</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; S</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; 2</span>}}$

Alternatively, newer measurements S ₁ (t _n ') and C ₁ (t _n ') from 9 can be used in place or combined with S ₁ (t _k ) and C ₁ (t _k ) use. The client also determines the quality of x _T (t _n ), eg by estimating the measurement variance σ _T ² .

11. Finally C combines the two bias measurements x _G and x _T into a final bias estimate x(t) taking into account their estimated quality. For example,

$\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; G</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>\frac{{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="T"\; filenames="HPA00001018085200011.png"\; state="\{\{state\}\}">T</figure-callout></span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}{{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="G"\; filenames="HPA00001018085200011.png"\; state="\{\{state\}\}">G</figure-callout></span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="T"\; filenames="HPA00001018085200011.png"\; state="\{\{state\}\}">T</figure-callout></span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; )</span>\frac{{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="G"\; filenames="HPA00001018085200011.png"\; state="\{\{state\}\}">G</figure-callout></span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}{{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="G"\; filenames="HPA00001018085200011.png"\; state="\{\{state\}\}">G</figure-callout></span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="T"\; filenames="HPA00001018085200011.png"\; state="\{\{state\}\}">T</figure-callout></span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}$

The client can finally adjust its clock according to C(t)←C(t)-x(t). The final bias estimate is derived formally as a maximum likelihood estimate from two independent Gaussian measurements x _G and x _T with variances σ _G ² and σ _T ² . In the practical case where GPS ^- derived measurements x _G are much more accurate than x _T in practice, _this simply _leads ^to x( _t )=x _G (t _i ). This procedure allows seamless or collision-free transition between two bias measurement schemes in the event that one bias measurement scheme fails and thus its variance increases.

12. Update times t _n , t _k , and t _i , and cycle from step 3 periodically.

Updates to time t _i (the time at which GPS time measurements and offset transfers are performed), time _tk (the time at which the server broadcasts the Sync message) and to time _tn (the time at which the two-way measurement exchange is performed) need not be synchronous. The latest available smoothed measurements should be used in the update algorithm. The update rates for these programs may be selected based on the availability of resources such as processor timing and network bandwidth. A higher rate increases the accuracy of the estimate, but at the expense of higher processing and communication load.

In order to accurately measure and correct the time offset x(t), it is also necessary to use successive time offset measurements to estimate the clock's frequency offset y(t). The basic assumption is (3), ie a linearly increasing clock skew, where a quadratic model is also conceivable.

The message delivery delays experience random jitter and outliers. Periodic repetition of the procedure allows the application of known smoothing algorithms to improve accuracy. Also, noise (jitter) variance can be estimated and other methods for evaluating measurement accuracy can be used. For example, a large value of the difference |C _k (t _n )−S _k (t _n )| is an outlier, indicating an isolated transmission problem affecting transmission delay, and should not be used to update the expected clock skew estimate. Recursive estimation algorithms that can interpolate between missing samples (such as temporary loss of GPS reception) or outliers of network delay can be applied to improve performance.

name list

1. Phasor Measurement Unit (PMU)

2. Network Control Center (NCC)

3. GPS satellite (GPS satellite)

4. Wide Area Communication Network

Claims (7)

1. method of be used for estimating the time deviation x between first and second static clock C, S, described static clock C, S are suitable for receiving full time signal G and are interconnected by communication network (4), and the method comprises:

-received the full time signal G (t be broadcasted with reference to (3) from the full time by the first and second static clock C, S _i), and at the first static clock place based on full time signal G (t _i) time of reception C (t _i), S (t _i) calculate between first and second static clock based on the clock jitter x that looks altogether _G(t _i);

-between first and second static clock C, S swap time prerequisite message, and at the first static clock place the transmission time S based on described message ₁(t _k), C ₂(t _n) and time of reception C ₁(t _k), S ₂(t _n) calculate the based on network clock jitter x between first and second static clock _T(t _n); And

-will be based on the clock jitter x that looks altogether _G(t _i) and based on network clock jitter x _T(t _n) combined with deviation x estimated time,

Wherein, variable t _i, t _kAnd t _nRefer to t constantly _i, t _kAnd t _n

2. method according to claim 1 is characterized in that comprising:

-estimate for based on the clock jitter x that looks altogether with network _G(t _i), x _T(t _n) quality σ _G, σ _TAnd

-pass through based on estimated quality σ _G, σ _TThe weighted average of calculating x makes up described two clock jitters.

3. method according to claim 1, is characterized in that, to based on the clock jitter x that looks altogether _G(t _i) calculating comprise:

-will comprise full time signal G (t by the second static clock in prerequisite mode of non-time _i) at the time of reception S (t at the second static clock place _i) message send to the first static clock.

4. method according to claim 1 is characterized in that comprising:

-upgrade independently and repeatedly based on the clock jitter x that looks altogether _G(t _i) and based on network clock jitter x _T(t _n) calculating.

5. method according to claim 1, it is characterized in that, the first static clock C is the clock of the phasor measurement unit (PMU) of wide area monitoring system, and the second static clock S is the clock of the network control center (NCC) of this wide area monitoring system.

6. method according to claim 5, is characterized in that, to based on network clock jitter x _T(t _n) calculating comprise

-will comprise that the prerequisite message of time of PMU phasor data sends to the second static clock S from the first static clock C.

One kind estimated time deviation equipment, it has static the first static clock C and is suitable for receiving full time signal G and exchanges messages by communication network (4), this equipment comprises:

-be used for based on full time signal G (t _i) the time of reception S of the position of the first and second static clock C, S ( _t), C (t _i) calculate between the first static clock C and the second static clock S based on the clock jitter c that looks altogether _G(t _i) device;

-for the transmission time S based on prerequisite message of time exchanged between the first and second static clock C, S ₁(t _k), C ₂(t _n) and time of reception C ₁( _tk), S ₂(t _n) calculate the based on network clock jitter x between the first static clock C and the second static clock S _T(t _n) device; And

-being used for will be based on the clock jitter x that looks altogether _G(t _i) and based on network clock jitter x _T(t _n) combined estimating the device of the time deviation x between the first static clock C and the second static clock S,

Wherein, variable t _i, t _kAnd t _nRefer to t constantly _i, t _kAnd t _n

Images